Is this a valid method for GO functional annotation?

[carojasq]

Hi all,

I am working on the functional annotation of a de novo assembled transcriptome. This is my procedure:

• BLASTX search against Refseq non-redundant protein database.
• Top hit selection from Blast search.
• Top hit association bettwen RefSeq id and GO with a idmapping table provided by Georgetown university: ftp://ftp.pir.georgetown.edu/databases/idmapping/. (A transcript can be associated with several GO terms).
• For each GO term associated with a contig I search the parent of the term in AmiGO browser, then i count the coincidences and generate a pie graph for it.

Is this a good way to do GO functional annotation? Would you recommend me another protocol to do this?

[Gopo]

I cannot comment on the final step but have performed steps 1-3 using the tool GOanna http://www.agbase.msstate.edu/cgi-bin/tools/GOanna.cgi and then used the resulting GO slim file and custom python scripts to develop probes targeting specific physiological processes. I would, however, advise you to be careful of the cutoff e-value you use for keeping BLASTX results, as you will likely not examine all of the matches by eye (i.e., "low" value matches even from the best result might be garbage).

[TiborNagy]

We have used Trinotate (part of the Trinity package) to annotate transcriptome.

[sepro]

We developed an on-line tool to do just that: TRAPID

Paper: http://genomebiology.com/2013/14/12/R134
Link: http://bioinformatics.psb.ugent.be/webtools/trapid/